Scientists apply deep learning to X-ray diffraction — ScienceDaily

X-ray diffraction (XRD) is an experimental technique for discerning the atomic structure of a material by irradiating it with X-rays at different angles. Essentially, the intensity of the reflected X-rays becomes high at specific irradiation angles, producing a pattern of diffraction peaks. An XRD acts like a fingerprint for a material because each substance produces a unique pattern. In research and development, changes in XRD are used to identify positions and amounts of additional elements that need to be added to fine-tune a material to help improve a desired functional property, say energy storage efficiency in batteries.

But the peak changes in XRD are barely discernible to humans. This makes it difficult to determine the characteristics and relevance of different peaks for material characterization. To this end, a group of Japanese researchers, led by Professor Ryo Maezono from the Japan Advanced Institute of Science and Technology (JAIST), applied a Deep Learning technique called “auto-encoder” to the problem of finding hidden regularities in XRD that could help to accelerate the development of new functional materials. The research group also included associate professor Kenta Hongo and assistant professor Kousuke Nakano from JAIST. Their work has been published in Advanced theory and simulations.

Explaining the basics of the autoencoding technique, Prof. Maezono says: “The autoencoding technique captures data features by expressing them as points on a two-dimensional plane (feature space). Based on their dispersion, the points are grouped into coarse-grained information. The autoencoder compresses the data dimension and can effectively capture the multifaceted XRD the pattern analysis in a two-dimensional plane.”

Using a neural network, the researchers applied the autoencoder to 150 XRD patterns of magnetic alloys with different concentrations. In feature space, each XRD is projected onto a single point. These points form clusters, in which similar materials with similar concentrations of constituents are placed closer together. Thus, the distance between the points in the feature space allows estimation of the concentration of a given sample alloy. This also enables fine-tuning of alloys by indirectly identifying the XRD peaks that change when new elements are added to an alloy or its constituents are changed.

The researchers further proposed a new application of the function space. When a peak of interest is masked on the original XRD pattern, the point shifts on the feature space. The magnitude of the shift helps distinguish how relevant a peak is to capture the properties of a material. Using this technique, the researchers were able to identify which peak is actually relevant to pay attention to in order to estimate the amount of doping, etc. — something that could not have been predicted by a human but was revealed using Deep Learning.

The researchers also proposed the application of auto-encoders for the generation of artificial XRD patterns by interpolating existing ones to handle small changes in alloy compositions. The approach would generate reasonable datasets and avoid computationally expensive ab initio simulations.

“The results of this research are not limited to XRD peak patterns. Rather, they provide a general Deep Learning technique that can be used to extract properties from material science data. Its framework can find hidden regularities in nature that are not identifiable by humans and are expected to work as a powerful tool for batch discovery through data science,” says Prof. Maezono.

The application of the described auto-encoder could accelerate the development of materials with high efficiency, low cost and low environmental impact, ushering in a new era of Deep Learning-based materials science research.

Story source:

Materials provided by Japan Advanced Institute of Science and Technology. Note! Content can be edited for style and length.

Leave a Reply

Your email address will not be published. Required fields are marked *